test(interop): required cross-vendor typed-data CI gate (Cyclone + Fast DDS) - #36
Closed
SandraK82 wants to merge 4 commits into
Closed
test(interop): required cross-vendor typed-data CI gate (Cyclone + Fast DDS)#36SandraK82 wants to merge 4 commits into
SandraK82 wants to merge 4 commits into
Conversation
Adds tests/interop/ci/: a vendor-agnostic runner that exchanges typed Robot samples over live DDS/RTPS between ZeroDDS and CycloneDDS / Fast DDS in both directions, asserting endpoint match AND decoded delivery (SPDP discovery alone is never a pass). The classifier distinguishes PASS, EXPECTED_NEGATIVE, PRODUCT_FAIL, TIMEOUT and SETUP_FAIL; machine-readable JSON + human summary + always-uploaded per-cell logs. - Pinned vendor installs (install_cyclone.sh / install_fastdds.sh); no floating latest images. - CycloneDDS cells retain the #29 XTypes semantics incl. the visible final-vs-appendable negative case. - Fast DDS client (fastdds_robot) generated from robot.idl by Fast-DDS-Gen. - Harness self-tests (classifier, exact-PID cleanup) run vendor-free in the interop-harness job. - ZeroDDS reader/writer + Cyclone client reused from #29, domain parameterised via ZERODDS_DOMAIN for per-cell isolation. Base main 80663c0. Vendor-live cells are validated by the public Linux CI in this PR.
Neither vendor is apt-installable on the hosted ubuntu-24.04 runner
("Unable to locate package libcyclonedds-dev") and eProsima publishes no
binary GitHub release asset, so both are now built from pinned git tags:
CycloneDDS 0.10.5 (+ matching cyclonedds-python 0.10.5), Fast CDR v2.2.4 +
Fast DDS v3.1.0 (bundled foonathan_memory) + Fast-DDS-Gen v4.0.1.
Deterministic, no floating latest, no unverified installer.
…y for Fast DDS (Refs #28) - The ZeroDDS interop reader/writer domain must be i32 (DomainParticipantFactory::create_participant takes DomainId = i32); the ZERODDS_DOMAIN parse used u32 and failed to compile (E0308), turning every Cyclone cell into SETUP_FAIL. - Fast DDS: build foonathan_memory_vendor + Fast CDR as separate pinned installs instead of -DTHIRDPARTY (a --depth 1 clone has no submodules, so find_package(foonathan_memory) failed).
…#28) The all-core Fast DDS source build (XTypes/TypeObject TUs are memory-heavy) peaked past the hosted-runner limit and the runner was SIGTERM'd twice at ~54% (exit 143) on the unchanged SHA — a resource limit, not transient infra. Cap the build to 2 jobs (CMAKE_BUILD_PARALLEL_LEVEL + gradle --max-workers) so the peak fits, add actions/cache keyed on the pinned versions so the stack builds once, and set a 60-minute job timeout for a clean failure mode. Cyclone cell is already green (both directions + the #29 XTypes cases).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Draft toward closing #28. Builds on #29 (
interop/cyclone-xtypes-27/) as prior foundation.Base
main@80663c06a87b629dfdcee8b5ba42201a971e104d(Merge PR #35 — the #27 multi-locator work; verified present:Vec<Locator>participant locators,if-addrs).Missing gate this addresses
The repo's cross-vendor interop is manual/gated (#29 is opt-in and Cyclone-only) or SPDP-discovery-only with floating
latestimages. There is no required public CI cell that exchanges typed user data with the open-source vendors in both directions.What this adds —
tests/interop/ci/A vendor-agnostic runner that, per vendor and per direction, exchanges typed
Robotsamples over live DDS/RTPS and asserts endpoint match AND decoded sample delivery (SPDP discovery alone is never a pass).interop_result.py— pure classifier: counts →PASS/EXPECTED_NEGATIVE/PRODUCT_FAIL/TIMEOUT/SETUP_FAIL(distinguishes an expected-negative from a product bug, a timeout and a setup failure).run_vendor.sh— orchestrates every cell: unique DDS domain per cell, exact-PID cleanup trap (never a broadpkill), boundedtimeouts,<vendor>-result.json+ human summary.lib.sh— shared exact-PID cleanup (unit tested).install_cyclone.sh/install_fastdds.sh— pinned vendor installs (nolatest); official packages/releases.fastdds/— Fast DDSRobotpub/sub client, type generated by Fast-DDS-Gen fromrobot.idl..github/workflows/ci.yml— two jobs:interop-harness(vendor-free classifier/cleanup/shellcheck) andinterop(matrixcyclone,fastdds,fail-fast: false, artifactsif: always()).The ZeroDDS reader/writer and Cyclone Python client are reused from #29, parameterised by
ZERODDS_DOMAINfor per-cell isolation (default 100 keeps the #29 matrix intact).Required matrix
CycloneDDS and Fast DDS, both directions, typed samples > 0, decode errors = 0 on the forward leg. CycloneDDS retains the #29 XTypes cases including the visible final-vs-appendable XCDR2 negative case (matched + zero decoded + decode errors surfaced via
take()), not a silent timeout.Validation
Portable layers, run locally on macOS (green):
python -m pytest tests/interop/ci/tests/test_interop_result.py -q→ 16 passed.bash tests/interop/ci/tests/test_cleanup.sh→ exact-PID cleanup spares an unrelated process.bash tests/interop/ci/run_vendor.sh fastdds <out>with the client absent → singleSETUP_FAILcell,result.json, exit 1.shellcheck -S warningclean;rustfmt --checkclean;yaml.safe_loadOK.Vendor-live cells (CycloneDDS + Fast DDS typed exchange, both directions) run on the hosted Linux CI in this PR — this host is macOS and cannot run the Linux vendor stacks, so per the contribution rules the public CI run is the evidence. This PR is a draft until those cells are green; the deliberate zero-sample mutation check will be exercised before it is marked ready.
Non-goals
RTI Connext in required CI, the full vendor×vendor matrix, DDS Security / SHM / TCP / TSN / WAN / soak / latency benchmarking, and fixing unrelated interop defects found while building the gate (filed separately).
Compatibility
No production/library code changes. The only edits outside
tests/interop/ci/make the #29 interop reader/writer and Cyclone client honourZERODDS_DOMAIN(default 100) — behaviour-preserving.Refs #28. Prior foundation: #29.